Analyze research~40 min

Analyze Research: Audit a Dataset You Did Not Collect

Early preview draft — not yet practitioner or accessibility reviewed.
Goal
Practice the due diligence that separates using data from being used by it: before a dataset informs a design decision, someone has to establish where it came from and what it can't say.
Scenario
Pick one public dataset or published survey a design team might actually cite (a design-industry salary survey, a public usage report, a government statistics table). Audit it before anyone is allowed to quote it in a design review.
Audience
A teammate about to put one of its numbers on a slide, who needs to know exactly how far the number can be trusted.
Constraints
The dataset must be real and public — link it. You are auditing, not analyzing: no charts of the data itself, only an assessment of it as evidence.
Deliverable
A one-page audit: who collected it, how, and when; who is IN the sample and who is missing; what consent/privacy posture the collection had; two specific claims the data CAN support and two it cannot; and a verdict — cite freely, cite with caveats, or don't cite.
Counts on your evidence profile as
A completed Skill Lab with its rubric and your deliverable
Accessibility requirement
If the dataset involves people, the audit must address whether they could reasonably expect this use of their data — that question is part of the craft, not an add-on.
AI policy
AI use is allowed in a limited way for this lab — see details below. AI may summarize the dataset's documentation; every provenance fact in your audit must be verified against the source itself, because documentation summaries are exactly where fabrication hides.
Rubric
  • Provenance is established: Collector, method, date, and sample are stated from the source's own documentation — with a flag where the documentation is silent, not a guess filling the gap.
  • The missing people are named: The audit identifies at least one population the sample under-represents and one claim that gap invalidates.
  • The verdict follows the evidence: The can/cannot-support claims connect visibly to the verdict — a reader could disagree, but not call it arbitrary.
Resources

Skills: Research Synthesis, Analytics

Relevant to: Product Designer, UX Researcher

Your work

Sign in to start this lab.